NCJ Number
78889
Editor(s)
R P Blanc
Date Published
1978
Length
55 pages
Annotation
This report identifies the approaches and techniques now practiced for detecting and correcting malperformance as it occurs in computer information systems.
Abstract
The work enumerates 67 items of current practice which prevent computer malperformance. In the development of the report, 65 persons known to be interested in the subject were contacted for their inputs. Correspondence and telephone calls elicited responses from 51 of these individuals. In addition, two common themes recurred throughout the literature search: error detection is costly, and perfection in data processing systems cannot be obtained for a reasonable price. Therefore, the report emphasizes that the system designer should pursue only the critical errors and design the system to limit the consequence of error. The report is intended for use by the systems designer who, using stock commercial hardware and software, is creating a system which will tax the available hardware, software, or staffing skills, and for the manager who wishes to chronicle the deficiencies in an existing system prior to improvement. Categories of ideas for improvement include data processing systems analysis, scientific systems analysis, implementation, and environment. The report suggests that when updating a data record during a file processing application, one should locally structure the application code so that it does not make any changes to the record until all tests have been passed and the transaction can be posted. This technique eliminates the need to reconstruct the original record in the event of error. Additional suggestions include peer review of the design effort, careful design of audit trials, and special attention to the mathematics of processing. Seven references, a glossary, a list of individuals contacted, and a subject index are appended.